Timeslot allocation for waiting list control

TL;DR

This paper develops and tests decision-making methods for hospital timeslot allocation to reduce patient wait times and optimize capacity, using Markov decision processes and simulation on real hospital data.

Contribution

It introduces a novel approach combining MDP modeling, approximate solution techniques, and a hybrid static-dynamic allocation method for hospital scheduling.

Findings

All proposed methods outperform static allocation.

The linear programming approach yields the best results.

Allocating 60% of timeslots statically and 40% dynamically is optimal.

Abstract

As pressure on the healthcare system increases, patients that require elective surgery experience longer access times to pre- and post-operative appointments and surgery. Hospitals can control their waiting lists by allocating timeslots to different types of appointments. To allow appointments to be planned timely, this allocation is decided several weeks in advance. However, the consequences of the timeslot allocation are uncertain, as not all patients follow the same treatment pathway. Furthermore, as these planning decisions are made in advance, they are based on an uncertain prediction of future waiting lists. We aim to develop methods that support hospitals in timeslot allocation to reduce access times for patients and ensure that all available capacity is used. The problem is modelled as a Markov decision process (MDP). As the state space is very large, we use least-squares policy iteration to find an approximate solution, formulate an (integer) linear program to solve a deterministic variant of the MDP, and investigate several decision rules. The solution methods are tested on a case study at the Sint Maartenskliniek, a hospital in the Netherlands. Based on a simulation study, we find that all methods improve on the currently used static allocation method, with the (integer) linear program leading to the best results. However, the performance deteriorates with the number of weeks the hospital plans ahead. To counter this, we propose a method in which a percentage of timeslots is statically allocated far in advance, and the remaining timeslots are allocated when enough information is available to effectively deal with variability. For the case study, we find that statically allocating 60% of the timeslots and dynamically allocating the remainder 6 weeks in advance provides the best results in terms of meeting access time targets and efficient resource utilization.